W3C home > Mailing lists > Public > www-tag@w3.org > May 2009

Re: Comment on XSD 1.1

From: Rick Jelliffe <rjelliffe@allette.com.au>
Date: Thu, 14 May 2009 04:40:51 +1000
Message-ID: <4A0B1433.9020602@allette.com.au>
To: noah_mendelsohn@us.ibm.com
CC: www-tag@w3.org, www-xml-schema-comments@w3.org
noah_mendelsohn@us.ibm.com wrote:
> Rick, 
>
> A minute ago I commented on this thread as someone who has been active in 
> the development of XML Schema.  Here I speak as TAG chair.
>   
Thanks for those kind words. It is very difficult to make these kind of 
comments without seeming
strident or opinionated or insulting, and I hope none of these are true.

> Rich Jelliffe writes:
>
>   
>> The appropriate channel for escalation in this case, as I understand
>> it, is the TAG.
>>     
>
> In short, I'm not convinced that's formally true, but I'm certainly glad 
> to try and get the TAG to play a constructive role, particularly if other 
> TAG members are inclined to engage in this issue.  I believe the process 
> requires that the TAG vote on whether to undertake this as an issue, and I 
> will schedule the appropriate vote shortly.
>   
All I can do is try. And I appreciate that this comes at perhaps a 
strange time near the end of a
very lengthy development effort. But if not now, when?
> In more detail, I think the pertinent issues with respect to W3C process 
> are:
>
> * A W3C Working Group has taken a draft specification to Candidate 
> Recommendation status.  You as a member of the community have expressed a 
> concern that I would summarize as:  the design is deeply flawed and should 
> not go forward as a W3C Recommendation.
>   
Yes, as long as "design" is not understood merely as "components and 
outcomes". The flaw is one
of failing to meet what I sumbit are probably minimum standard or core 
requirements.
My belief is that this comes from imposing on all schemas an 
inappropriately
grandiose theoretical mechanism (complex type derivation) that remains 
perfectly useable for
at least  a significant niche of potential users. However, the XSD WG 
has not considered
a type-derivation-free, relaxed version of schemas, nor would it do so 
off its own bat. (In fact
it will not consider any subsetting or simplification ideas at all, 
apart from simplifying rules that
are so complex that even those who can work with XSD cannot use them.)

And my solution is layering and a relaxed, broader,  simpler, 
derivation-free layer suitable
for simple use, syntactically compatible with XSD to the  greatest 
extent. Build XSD 1.1
on top of that.

However, it may be that the TAG could decide that XSD 1.0 has, in the 
light of experience,
proved itself deeply flawed and failing to meet core requirements, and 
that the program of
word or the dynamics of the the XSD WG need to be electro-shocked into a 
more useful
state, without formally recommend my particular solution. I don't think 
anyone has come up
with a better solution, but the XSD WG has clearly not tried.

(However, RELAX NG, whose
semantics I suggest would be appropriate as the new core, with modified 
XSD-namespace
syntax, is well thought out, implemented, standardized and has more 
complete theoretical
characterization than XSD. So it should not be thought I am suggesting a 
whole new development
effort or branching out into the unknown. You could perhaps see my plan 
as making
RELAX NG the simple base schema language, then using it to simplify the 
XSD 1.n specification,
retaining the XSD syntax as much as possible. For example, the core 
would follow RELAX NG in
only allowing cardinality of 0, 1 or unbounded, but it would still 
express this using XSD's minOccurs
and maxOccurs.)
> * The normal W3C Process involves processing of comments by the schemas 
> working group. 
There are two problems with this. The minor one is that I think the time 
for public comment is over.

The major one is that the WG had already decided it would not address 
the issues I have re-raised
and did XSD 1.1 instead, as Michael wrote. And the public comments to 
the WG on XSD 1.1
could only in practice address specific issues, not the broad picture.

So I believe it is utterly useful to attempt to use the XSD WG to raise 
this issue.

> Sorry for all the process stuff, but since you said that the TAG is the 
> appropriate channel for escalation, I thought it was worth rechecking the 
> groundrules.
>
>   
No, I am grateful to you for responding. It am aware that it must cheese 
off everyone on the WG who
has been diligently working on 1.1 for so long with such little result 
or public warmth. I get nice comments
about Schematron all the time: I don't think I have ever had a nasty 
one. Murata seems pretty happy
with feedback on RELAX NG too. Public frustration is not part of  the 
territory for schema languages
that schema standards committee people must lump, or acclimatize 
themselves to, it is a sign that
something is amiss.
> Here's my net:  Rick has, at least informally, asked the TAG to consider 
> his concerns with XML Schema.  Consistent with all the process stuff 
> quoted above, I hereby ask the TAG to consider whether it wishes to 
> undertake work on this issue, and if so how to prioritize relative other 
> things.  I will take the appropriate vote on an upcoming teleconference. 
> If the TAG decides not to get involved at this time, then there is also 
> the opportunity for Tim to ask the TAG's advice during the approval 
> process for XSD 1.1.
>  
Thanks Noah. Please consider it formally asked, if I have that privilege 
as a member of the public.

Cheers
Rick Jelliffe
Received on Wednesday, 13 May 2009 18:41:38 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 26 April 2012 12:48:13 GMT